About Knot
Knot is a remote MCP server for carrying project context across AI clients, aimed at individuals and teams who use multiple AI tools. It records facts, decisions, and goals, and supports handoffs to continue tasks across agents.
Key features
- 01Retrieve current, historical, and trajectory context
- 02Recover sessions across AI clients
- 03Save facts, decisions, constraints, and goals
- 04Create and check task handoffs between agents
- 05Switch between personal and shared team scopes
Use cases
- 01Continue project context across different AI clients
- 02Resume work from another AI session
- 03Record and update project decisions and constraints
- 04Hand a multi-step task to another agent
MCP support
Knot provides a remote MCP endpoint and uses OAuth 2.1 with PKCE for authentication. It exposes tools for context retrieval and saving, session recovery, agent handoffs, and scope management.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://api.connectknot.com/mcpclaude mcp add --transport http com-connectknot-knot https://api.connectknot.com/mcp{"mcpServers":{"com-connectknot-knot":{"type":"http","url":"https://api.connectknot.com/mcp"}}}Open in Cursor{"servers":{"com-connectknot-knot":{"type":"http","url":"https://api.connectknot.com/mcp"}}}Open in VS Code{"mcpServers":{"com-connectknot-knot":{"type":"http","url":"https://api.connectknot.com/mcp"}}}{"mcpServers":{"com-connectknot-knot":{"serverUrl":"https://api.connectknot.com/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMEKnot MCP Server
One shared brain for all your AIs.
Knot (https://api.connectknot.com/mcp) is a remote Model Context Protocol (MCP) server with OAuth 2.1 + PKCE authentication. Connect ChatGPT, Claude, Cursor, VS Code, Cline, Windsurf, and Gemini so your decisions, project state, and vetted context carry across every AI you use without starting over.
- Website: https://connectknot.com
- Developer Quickstart: https://connectknot.com/developers/quickstart
- Remote MCP Endpoint:
https://api.connectknot.com/mcp(streamable-http) - Official MCP Registry:
com.connectknot/knot&io.github.ConnectKnot/knot
1-Click Install
- Cursor:
cursor://anysphere.cursor-deeplink/mcp/install?name=knot&config=eyJ1cmwiOiJodHRwczovL2FwaS5jb25uZWN0a25vdC5jb20vbWNwIn0%3D - VS Code:
vscode:mcp/install?%7B%22name%22%3A%22knot%22%2C%22type%22%3A%22http%22%2C%22url%22%3A%22https%3A%2F%2Fapi.connectknot.com%2Fmcp%22%7D - Claude Code CLI:
claude mcp add --transport http knot https://api.connectknot.com/mcp
Standard mcp.json Configuration (OAuth 2.1)
For Cursor, VS Code, Windsurf, Cline, or Claude Desktop:
{
"mcpServers": {
"knot": {
"type": "http",
"url": "https://api.connectknot.com/mcp"
}
}
}
When your AI client connects for the first time, it discovers https://api.connectknot.com/.well-known/oauth-protected-resource and opens your browser to sign in and approve access on connectknot.com.
Core MCP Tools
| Tool | Description |
|---|---|
get_relevant_context | Point-in-time factual retrieval of active, historical, or trajectory context across your personal or shared project Knots. |
get_context_recovery | Full cross-AI session recovery to pick up where another AI left off. |
propose_context / batch_propose_context | Save durable facts, decisions, constraints, and goals with explicit supersession lineage (updates_ref / supersedes_context_id). |
create_handoff / check_pending_handoffs | Hand off structured multi-step tasks between connected AI agents (e.g., Claude -> Cursor). |
knot_list_scopes / knot_set_active_scope | Switch between your Personal Knot and collaborative Team Shared Knots. |
License
MIT © Knot
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.